Struct Prediction 

pub struct Prediction {
    pub id: String,
    pub claim_text: String,
    pub target_findings: Vec<String>,
    pub predicted_at: String,
    pub resolves_by: Option<String>,
    pub resolution_criterion: String,
    pub expected_outcome: ExpectedOutcome,
    pub made_by: String,
    pub confidence: f64,
    pub conditions: Conditions,
    pub expired_unresolved: bool,
}

v0.34: Prediction as a first-class kernel object.

A Prediction is a falsifiable claim about a future observation, scoped to one or more existing findings, made by a registered actor at a known timestamp, with an explicit resolution criterion and (typically) a deadline. Resolutions arrive later as Resolution records that close out the prediction by recording what actually happened.

Predictions are the kernel’s epistemic accountability layer. Other parts of the substrate describe what is believed today; predictions describe what is expected and let the substrate score, over time, how well each actor’s beliefs track reality. Calibration records (Brier, log score, hit rate) are derived from the resolved subset.

vpred_<id> is content-addressed over claim_text + made_by + predicted_at + resolution_criterion. Two predictions with the same prose but different actors or different criteria are distinct kernel objects.

Fields

id: String

vpred_<16hex>, content-addressed.

claim_text: String

The falsifiable prediction itself, in plain prose.

target_findings: Vec<String>

Existing vf_* findings whose truth this prediction depends on. May be empty for predictions that don’t tie back to a specific frontier claim.

predicted_at: String

RFC 3339 timestamp of when the prediction was made. Goes into the content-address preimage so re-asserting the same prose at a later date produces a distinct record.

resolves_by: Option<String>

RFC 3339 deadline for resolution. None means open-ended; a concrete date is strongly preferred for calibration scoring.

resolution_criterion: String

Unambiguous prose that says “we’ll know this resolved when X.” Goes into the content-address preimage so the same prose with a different criterion is a distinct record.

expected_outcome: ExpectedOutcome

Structured expectation: affirmed / falsified / quantitative / categorical. The resolver checks this against the observed outcome at resolution time.

made_by: String

Stable actor id of the predictor.

confidence: f64

Predictor’s prior belief in the expected outcome, on [0, 1]. Drives Brier scoring at resolution time.

conditions: Conditions

Conditions under which the prediction applies. Reuses the Conditions shape so model relevance, scope, etc., flow through.

expired_unresolved: bool

v0.40.1: True once the calibration runtime has marked this prediction as expired without an explicit Resolution. Set by calibration::expire_overdue_predictions when resolves_by is in the past. Pre-v0.40.1 frontiers omit the field; loading is backward-compatible. An expired prediction does not become a resolved prediction — it is closed without contributing to Brier or log scoring (calibration tracks it as a separate n_expired count so the predictor still answers for the missing commitment).

Implementations

impl Prediction

pub fn content_address( claim_text: &str, made_by: &str, predicted_at: &str, resolution_criterion: &str, expected_outcome: &ExpectedOutcome, ) -> String

Compute the content-addressed ID per v0.34 spec: SHA-256(normalize(claim_text) | made_by | predicted_at | normalize(resolution_criterion) | expected_outcome.canonical()). Returns first 16 hex chars prefixed with “vpred_”.

pub fn new( claim_text: impl Into<String>, target_findings: Vec<String>, predicted_at: Option<String>, resolves_by: Option<String>, resolution_criterion: impl Into<String>, expected_outcome: ExpectedOutcome, made_by: impl Into<String>, confidence: f64, conditions: Conditions, ) -> Self

Construct a new Prediction. predicted_at defaults to “now” in RFC 3339 if not supplied.

Trait Implementations

impl Clone for Prediction

fn clone(&self) -> Prediction

Returns a duplicate of the value.

fn clone_from(&mut self, source: &Self)

Performs copy-assignment from source.

impl Debug for Prediction

fn fmt(&self, f: &mut Formatter<'_>) -> Result

Formats the value using the given formatter.

impl<'de> Deserialize<'de> for Prediction

fn deserialize<__D>(__deserializer: __D) -> Result<Self, __D::Error>

where __D: Deserializer<'de>,

Deserialize this value from the given Serde deserializer.

impl Serialize for Prediction

fn serialize<__S>(&self, __serializer: __S) -> Result<__S::Ok, __S::Error>

where __S: Serializer,

Serialize this value into the given Serde serializer.